Producer phase data and/or model available for download?

[JViggiani]

Hello!

I have downloaded this repository and I'm interested in training this model to play around with.

However, the LLB producer phase I estimate will take 182 days of continuous training for me to just synthesise the text embeddings before training the model.

Is the producer.py synthetic data available for download anywhere?

And further is the stage 1 and stage 2 trained model(s) available to download anywhere?

Thank you!

[smw1996]

Dear user,

Thank you for your interest in our project. You can find the trained models in our README file; they are available for direct download via this link: https://huggingface.co/Hualouz/Qformer/tree/main.

Regarding the synthetic data generated by producer.py, we are currently evaluating the possibility of making it open-source. If we determine it is feasible to do so, we will strive to make it available as soon as possible so that community members like yourself can more easily engage in experimentation and research.

We appreciate your patience and interest once again. Should there be any updates, we will be sure to inform the community promptly.

Best regards!

[xxxl123]

Could you please tell me approximately how much time each step takes on the same device, V100?

[rajnish159]

is this model producer.py available for download?? and can i skip this step??, because i dont have much GPU. and total paper count: 169343 , can i reduce the paper count??